Cygwin: x86_64: import latest NetBSD bcopy.S
Tweak slightly to allow implementing entire {w}mem{p}{cpy,move}
family:
Add WIDE macro processing for wmem* and POST macro processing for
memp* functions.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
This commit is contained in:
Corinna Vinschen
parent
b09617a828
commit
43743ed754
|
|
@ -52,9 +52,15 @@ TEST_LIB_NAME=libcygwin0.a
|
||||||
# These objects are included directly into the import library
|
# These objects are included directly into the import library
|
||||||
if TARGET_X86_64
|
if TARGET_X86_64
|
||||||
TARGET_FILES= \
|
TARGET_FILES= \
|
||||||
|
x86_64/bcopy.S \
|
||||||
x86_64/memchr.S \
|
x86_64/memchr.S \
|
||||||
x86_64/memcpy.S \
|
x86_64/memcpy.S \
|
||||||
x86_64/memset.S
|
x86_64/memmove.S \
|
||||||
|
x86_64/mempcpy.S \
|
||||||
|
x86_64/memset.S \
|
||||||
|
x86_64/wmemcpy.S \
|
||||||
|
x86_64/wmemmove.S \
|
||||||
|
x86_64/wmempcpy.S
|
||||||
endif
|
endif
|
||||||
|
|
||||||
LIB_FILES= \
|
LIB_FILES= \
|
||||||
|
|
|
||||||
|
|
@ -0,0 +1,192 @@
|
||||||
|
/*-
|
||||||
|
* Copyright (c) 1990 The Regents of the University of California.
|
||||||
|
* All rights reserved.
|
||||||
|
*
|
||||||
|
* This code is derived from locore.s.
|
||||||
|
*
|
||||||
|
* Redistribution and use in source and binary forms, with or without
|
||||||
|
* modification, are permitted provided that the following conditions
|
||||||
|
* are met:
|
||||||
|
* 1. Redistributions of source code must retain the above copyright
|
||||||
|
* notice, this list of conditions and the following disclaimer.
|
||||||
|
* 2. Redistributions in binary form must reproduce the above copyright
|
||||||
|
* notice, this list of conditions and the following disclaimer in the
|
||||||
|
* documentation and/or other materials provided with the distribution.
|
||||||
|
* 3. Neither the name of the University nor the names of its contributors
|
||||||
|
* may be used to endorse or promote products derived from this software
|
||||||
|
* without specific prior written permission.
|
||||||
|
*
|
||||||
|
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
||||||
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||||
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
||||||
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
||||||
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
||||||
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
||||||
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
||||||
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
||||||
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
||||||
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
||||||
|
* SUCH DAMAGE.
|
||||||
|
*/
|
||||||
|
|
||||||
|
#include <machine/asm.h>
|
||||||
|
|
||||||
|
#if defined(LIBC_SCCS)
|
||||||
|
RCSID("$NetBSD: bcopy.S,v 1.5 2014/03/22 19:16:34 jakllsch Exp $")
|
||||||
|
#endif
|
||||||
|
|
||||||
|
/*
|
||||||
|
* (ov)bcopy (src,dst,cnt)
|
||||||
|
* ws@tools.de (Wolfgang Solfrank, TooLs GmbH) +49-228-985800
|
||||||
|
*
|
||||||
|
* Hacked about by dsl@netbsd.org
|
||||||
|
*/
|
||||||
|
|
||||||
|
#ifdef MEMCOPY
|
||||||
|
#ifdef WIDE
|
||||||
|
#ifdef POST
|
||||||
|
ENTRY3(wmempcpy)
|
||||||
|
#else
|
||||||
|
ENTRY3(wmemcpy)
|
||||||
|
#endif
|
||||||
|
#else
|
||||||
|
#ifdef POST
|
||||||
|
ENTRY3(mempcpy)
|
||||||
|
#else
|
||||||
|
ENTRY3(memcpy)
|
||||||
|
#endif
|
||||||
|
#endif
|
||||||
|
#define NO_OVERLAP
|
||||||
|
#else
|
||||||
|
#ifdef MEMMOVE
|
||||||
|
#ifdef WIDE
|
||||||
|
ENTRY3(wmemmove)
|
||||||
|
#else
|
||||||
|
ENTRY3(memmove)
|
||||||
|
#endif
|
||||||
|
#else
|
||||||
|
ENTRY3(bcopy)
|
||||||
|
#endif
|
||||||
|
#endif
|
||||||
|
#ifdef WIDE
|
||||||
|
shlq $1,%rdx /* cnt * sizeof (wchar_t) */
|
||||||
|
#endif
|
||||||
|
movq %rdx,%rcx
|
||||||
|
#if defined(MEMCOPY) || defined(MEMMOVE)
|
||||||
|
movq %rdi,%rax /* must return destination address */
|
||||||
|
#ifdef POST
|
||||||
|
addq %rdx,%rax /* + n */
|
||||||
|
#endif
|
||||||
|
mov %rdi,%r11 /* for misaligned check */
|
||||||
|
#else
|
||||||
|
mov %rsi,%r11 /* for misaligned check */
|
||||||
|
xchgq %rdi,%rsi /* bcopy() has arg order reversed */
|
||||||
|
#endif
|
||||||
|
|
||||||
|
#if !defined(NO_OVERLAP)
|
||||||
|
movq %rdi,%r8
|
||||||
|
subq %rsi,%r8
|
||||||
|
#endif
|
||||||
|
|
||||||
|
shrq $3,%rcx /* count for copy by words */
|
||||||
|
jz 8f /* j if less than 8 bytes */
|
||||||
|
|
||||||
|
lea -8(%rdi,%rdx),%r9 /* target address of last 8 */
|
||||||
|
mov -8(%rsi,%rdx),%r10 /* get last word */
|
||||||
|
#if !defined(NO_OVERLAP)
|
||||||
|
cmpq %rdx,%r8 /* overlapping? */
|
||||||
|
jb 10f
|
||||||
|
#endif
|
||||||
|
|
||||||
|
/*
|
||||||
|
* Non-overlaping, copy forwards.
|
||||||
|
* Newer Intel cpus (Nehalem) will do 16byte read/write transfers
|
||||||
|
* if %ecx is more than 76.
|
||||||
|
* AMD might do something similar some day.
|
||||||
|
*/
|
||||||
|
and $7,%r11 /* destination misaligned ? */
|
||||||
|
jnz 2f
|
||||||
|
rep
|
||||||
|
movsq
|
||||||
|
mov %r10,(%r9) /* write last word */
|
||||||
|
ret
|
||||||
|
|
||||||
|
/*
|
||||||
|
* Destination misaligned
|
||||||
|
* AMD say it is better to align the destination (not the source).
|
||||||
|
* This will also re-align copies if the source and dest are both
|
||||||
|
* misaligned by the same amount)
|
||||||
|
* (I think Nehalem will use its accelerated copy if the source
|
||||||
|
* and destination have the same alignment.)
|
||||||
|
*/
|
||||||
|
2:
|
||||||
|
lea -9(%r11,%rdx),%rcx /* post re-alignment count */
|
||||||
|
neg %r11 /* now -1 .. -7 */
|
||||||
|
mov (%rsi),%rdx /* get first word */
|
||||||
|
mov %rdi,%r8 /* target for first word */
|
||||||
|
lea 8(%rsi,%r11),%rsi
|
||||||
|
lea 8(%rdi,%r11),%rdi
|
||||||
|
shr $3,%rcx
|
||||||
|
rep
|
||||||
|
movsq
|
||||||
|
mov %rdx,(%r8) /* write first word */
|
||||||
|
mov %r10,(%r9) /* write last word */
|
||||||
|
ret
|
||||||
|
|
||||||
|
#if !defined(NO_OVERLAP)
|
||||||
|
/* Must copy backwards.
|
||||||
|
* Reverse copy is probably easy to code faster than 'rep movds'
|
||||||
|
* since that requires (IIRC) an extra clock every 3 iterations (AMD).
|
||||||
|
* However I don't suppose anything cares that much!
|
||||||
|
* The big cost is the std/cld pair - reputedly 50+ cycles on Netburst P4.
|
||||||
|
* The copy is aligned with the buffer start (more likely to
|
||||||
|
* be a multiple of 8 than the end).
|
||||||
|
*/
|
||||||
|
10:
|
||||||
|
lea -8(%rsi,%rcx,8),%rsi
|
||||||
|
lea -8(%rdi,%rcx,8),%rdi
|
||||||
|
std
|
||||||
|
rep
|
||||||
|
movsq
|
||||||
|
cld
|
||||||
|
mov %r10,(%r9) /* write last bytes */
|
||||||
|
ret
|
||||||
|
#endif
|
||||||
|
|
||||||
|
/* Less than 8 bytes to copy, copy by bytes */
|
||||||
|
/* Intel Nehalem optimise 'rep movsb' for <= 7 bytes (9-15 clocks).
|
||||||
|
* For longer transfers it is 50+ !
|
||||||
|
*/
|
||||||
|
8: mov %rdx,%rcx
|
||||||
|
|
||||||
|
#if !defined(NO_OVERLAP)
|
||||||
|
cmpq %rdx,%r8 /* overlapping? */
|
||||||
|
jb 81f
|
||||||
|
#endif
|
||||||
|
|
||||||
|
/* nope, copy forwards. */
|
||||||
|
rep
|
||||||
|
movsb
|
||||||
|
ret
|
||||||
|
|
||||||
|
#if !defined(NO_OVERLAP)
|
||||||
|
/* Must copy backwards */
|
||||||
|
81:
|
||||||
|
lea -1(%rsi,%rcx),%rsi
|
||||||
|
lea -1(%rdi,%rcx),%rdi
|
||||||
|
std
|
||||||
|
rep
|
||||||
|
movsb
|
||||||
|
cld
|
||||||
|
ret
|
||||||
|
#endif
|
||||||
|
|
||||||
|
#ifdef MEMCOPY
|
||||||
|
END(memcpy)
|
||||||
|
#else
|
||||||
|
#ifdef MEMMOVE
|
||||||
|
END(memmove)
|
||||||
|
#else
|
||||||
|
END(bcopy)
|
||||||
|
#endif
|
||||||
|
#endif
|
||||||
|
|
@ -1,131 +1,4 @@
|
||||||
/* These functions are almost verbatim FreeBSD code (even if the header of
|
/* $NetBSD: memcpy.S,v 1.1 2005/12/20 19:28:51 christos Exp $ */
|
||||||
one file mentiones NetBSD), just wrapped in the minimum required code to
|
|
||||||
make them work under the MS AMD64 ABI.
|
|
||||||
See FreeBSD src/lib/libc/amd64/string/bcopy.S */
|
|
||||||
|
|
||||||
/*-
|
#define MEMCOPY
|
||||||
* Copyright (c) 1990 The Regents of the University of California.
|
#include "bcopy.S"
|
||||||
* All rights reserved.
|
|
||||||
*
|
|
||||||
* This code is derived from locore.s.
|
|
||||||
*
|
|
||||||
* Redistribution and use in source and binary forms, with or without
|
|
||||||
* modification, are permitted provided that the following conditions
|
|
||||||
* are met:
|
|
||||||
* 1. Redistributions of source code must retain the above copyright
|
|
||||||
* notice, this list of conditions and the following disclaimer.
|
|
||||||
* 2. Redistributions in binary form must reproduce the above copyright
|
|
||||||
* notice, this list of conditions and the following disclaimer in
|
|
||||||
* the documentation and/or other materials provided with the
|
|
||||||
* distribution.
|
|
||||||
* 3. Neither the name of the University nor the names of its
|
|
||||||
* contributors may be used to endorse or promote products derived
|
|
||||||
* from this software without specific prior written permission.
|
|
||||||
*
|
|
||||||
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS''
|
|
||||||
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
|
|
||||||
* THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
|
|
||||||
* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR
|
|
||||||
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
|
||||||
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
|
||||||
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
|
||||||
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
|
|
||||||
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
||||||
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
|
|
||||||
* USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
|
|
||||||
* DAMAGE.
|
|
||||||
*/
|
|
||||||
|
|
||||||
.seh_proc _memcpy
|
|
||||||
_memcpy:
|
|
||||||
movq %rsi,8(%rsp)
|
|
||||||
movq %rdi,16(%rsp)
|
|
||||||
.seh_endprologue
|
|
||||||
movq %rcx,%rdi
|
|
||||||
movq %rdx,%rsi
|
|
||||||
movq %r8,%rdx
|
|
||||||
|
|
||||||
movq %rdx,%rcx
|
|
||||||
movq %rdi,%r8
|
|
||||||
subq %rsi,%r8
|
|
||||||
cmpq %rcx,%r8 /* overlapping? */
|
|
||||||
jb 1f
|
|
||||||
cld /* nope, copy forwards. */
|
|
||||||
shrq $3,%rcx /* copy by words */
|
|
||||||
rep movsq
|
|
||||||
movq %rdx,%rcx
|
|
||||||
andq $7,%rcx /* any bytes left? */
|
|
||||||
rep movsb
|
|
||||||
jmp 2f
|
|
||||||
1:
|
|
||||||
addq %rcx,%rdi /* copy backwards. */
|
|
||||||
addq %rcx,%rsi
|
|
||||||
std
|
|
||||||
andq $7,%rcx /* any fractional bytes? */
|
|
||||||
decq %rdi
|
|
||||||
decq %rsi
|
|
||||||
rep movsb
|
|
||||||
movq %rdx,%rcx /* copy remainder by words */
|
|
||||||
shrq $3,%rcx
|
|
||||||
subq $7,%rsi
|
|
||||||
subq $7,%rdi
|
|
||||||
rep movsq
|
|
||||||
cld
|
|
||||||
2:
|
|
||||||
movq 8(%rsp),%rsi
|
|
||||||
movq 16(%rsp),%rdi
|
|
||||||
ret
|
|
||||||
.seh_endproc
|
|
||||||
|
|
||||||
.globl memmove
|
|
||||||
.seh_proc memmove
|
|
||||||
memmove:
|
|
||||||
.seh_endprologue
|
|
||||||
movq %rcx,%rax /* return dst */
|
|
||||||
jmp _memcpy
|
|
||||||
.seh_endproc
|
|
||||||
|
|
||||||
.globl memcpy
|
|
||||||
.seh_proc memcpy
|
|
||||||
memcpy:
|
|
||||||
.seh_endprologue
|
|
||||||
movq %rcx,%rax /* return dst */
|
|
||||||
jmp _memcpy
|
|
||||||
.seh_endproc
|
|
||||||
|
|
||||||
.globl mempcpy
|
|
||||||
.seh_proc mempcpy
|
|
||||||
mempcpy:
|
|
||||||
.seh_endprologue
|
|
||||||
movq %rcx,%rax /* return dst */
|
|
||||||
addq %r8,%rax /* + n */
|
|
||||||
jmp _memcpy
|
|
||||||
.seh_endproc
|
|
||||||
|
|
||||||
.globl wmemmove
|
|
||||||
.seh_proc wmemmove
|
|
||||||
wmemmove:
|
|
||||||
.seh_endprologue
|
|
||||||
shlq $1,%r8 /* cnt * sizeof (wchar_t) */
|
|
||||||
movq %rcx,%rax /* return dst */
|
|
||||||
jmp _memcpy
|
|
||||||
.seh_endproc
|
|
||||||
|
|
||||||
.globl wmemcpy
|
|
||||||
.seh_proc wmemcpy
|
|
||||||
wmemcpy:
|
|
||||||
.seh_endprologue
|
|
||||||
shlq $1,%r8 /* cnt * sizeof (wchar_t) */
|
|
||||||
movq %rcx,%rax /* return dst */
|
|
||||||
jmp _memcpy
|
|
||||||
.seh_endproc
|
|
||||||
|
|
||||||
.globl wmempcpy
|
|
||||||
.seh_proc wmempcpy
|
|
||||||
wmempcpy:
|
|
||||||
.seh_endprologue
|
|
||||||
shlq $1,%r8 /* cnt * sizeof (wchar_t) */
|
|
||||||
movq %rcx,%rax /* return dst */
|
|
||||||
addq %r8,%rax /* + n */
|
|
||||||
jmp _memcpy
|
|
||||||
.seh_endproc
|
|
||||||
|
|
|
||||||
|
|
@ -0,0 +1,4 @@
|
||||||
|
/* $NetBSD: memmove.S,v 1.1 2005/12/20 19:28:51 christos Exp $ */
|
||||||
|
|
||||||
|
#define MEMMOVE
|
||||||
|
#include "bcopy.S"
|
||||||
|
|
@ -0,0 +1,5 @@
|
||||||
|
/* $NetBSD: memcpy.S,v 1.1 2005/12/20 19:28:51 christos Exp $ */
|
||||||
|
|
||||||
|
#define MEMCOPY
|
||||||
|
#define POST
|
||||||
|
#include "bcopy.S"
|
||||||
|
|
@ -0,0 +1,5 @@
|
||||||
|
/* $NetBSD: memcpy.S,v 1.1 2005/12/20 19:28:51 christos Exp $ */
|
||||||
|
|
||||||
|
#define MEMCOPY
|
||||||
|
#define WIDE
|
||||||
|
#include "bcopy.S"
|
||||||
|
|
@ -0,0 +1,5 @@
|
||||||
|
/* $NetBSD: memmove.S,v 1.1 2005/12/20 19:28:51 christos Exp $ */
|
||||||
|
|
||||||
|
#define MEMMOVE
|
||||||
|
#define WIDE
|
||||||
|
#include "bcopy.S"
|
||||||
|
|
@ -0,0 +1,6 @@
|
||||||
|
/* $NetBSD: memcpy.S,v 1.1 2005/12/20 19:28:51 christos Exp $ */
|
||||||
|
|
||||||
|
#define MEMCOPY
|
||||||
|
#define WIDE
|
||||||
|
#define POST
|
||||||
|
#include "bcopy.S"
|
||||||
Loading…
Reference in New Issue